There is probably no greater struggle in our world of tantalizing food porn and convenient food delivery apps than Healthy eating. Create more deliciousness nutritionally balanced meals it feels like a chore with too many extra steps: researching recipes, grocery shopping, cooking, and meal prep.
In the age of generative artificial intelligence, some Internet users are turning to chatbots to try to establish a better relationship with their customers. food. On social media forums like Reddit, users have shared their experiences using ChatGPT, trading prompts, hacks, and meal tips that are, for example, high in protein and low in carbohydrates Or “good for losing weight.” The same goes for TikTok where there is even more specific advicesuch as creating “hormonally balanced” meals for reproductive issues like polycystic ovary syndrome.
But using ChatGPT to take the guesswork out of living a healthy, nutritious lifestyle reveals an inconvenient truth about the state of nutritional science and how ChatGPT interprets scientific studies. Research reveals that poor science is hiding in algorithms, which could contribute to life-threatening risks, such as disordered eating habits or nutritional advice that could trigger serious health problems.
Trained on questionable information
Since its launch in 2022, ChatGPT has seen a multitude of uses, some very basic, such as ask the chatbot how to cook meth — to more pragmatic ones like give software and search engines a more human dimension Or improve scientific discovery.
The problem with ChatGPT is that while it is equipped with seemingly encyclopedic knowledge on a number of topics, that information is only as good as the data it was trained on. When it comes to nutrition and diet, there is a lot of information on the Internet that may not be scientifically valid or applicable to humans, says Lindsay Maloneregistered dietitian who practices integrative and functional dietetics at Case Western Reserve University.
The problem with ChatGPT is that while it is equipped with seemingly encyclopedic knowledge on a number of topics, that information is only as good as the data it was trained on. When it comes to nutrition and diet, there is a lot of information on the Internet that may not be scientifically valid or applicable to humans.
vgajic/E+/Getty Images
“There are many health and wellness blogs or websites that may contain statements that are not necessarily false. They may be based on animal data or smaller studies where we have evidence to move forward with a large-scale study, but we can’t really change our recommendations yet until we have seeing this proven on a larger human scale,” she explains. Reverse.
ChatGPT strives to include fruits and vegetables in every meal for diets like the Mediterranean Diet or Dietary Approaches to Stop Hypertension (also known as DASH). For example, on a Mediterranean diet – which the AI describes “emphasizing fresh produce, whole grains, healthy fats, lean proteins and moderate wine consumption” – it recommended a vegetarian omelette for the little one. -lunch, cooked with olive oil and toasted almonds with rosemary as a snack. For those following DASH, ChatGPT recommends avoiding saturated fats, sugars and sodium and instead calls for using “herbs and spices” to flavor a bowl of quinoa and vegetables at lunch or dinner. low-sodium soy sauce for a vegetable stir-fry.
“(A registered dietitian) would provide some general information: find out your goals, if there are any health issues or food allergies, or if a person is looking to change their health to a certain extent, such as their blood cholesterol levels or their body composition. »
However, a item 2023 published in the Journal of the Academy of Nutrition and Dietetics, found that specifying a diet did not mean the chatbot always took these dietary constraints into account. As an experiment, the authors of the article asked ChatGPT to provide an ideal diet for people with failing kidneys who are on dialysis. The AI provided a specific answer, such as reducing potassium and phosphorus intake and consulting a “kidney dietitian.” But when asked to create a week’s menu, the foods included, like spinach and avocado, were not optimal for dialysis patients and provided no warning, according to the authors.
Another Study 2023 published in the journal Nutrition explored using ChatGPT to create food allergy-friendly meals. The researchers focused on 14 food allergies, including gluten, eggs, fish, shellfish, tree nuts and dairy. Most of the menus created correctly by the chatbot omitted the specified food allergen. But for nut-free diets, ChatGPT included almond milk, one of the most serious and potentially fatal foods allergies.
Reinforce a poor diet
Perhaps the most serious danger of ChatGPT advice is that it has the potential to reinforce or encourage eating disorders.
This is not a new source of concern regarding generative AI. Last summer, the National Eating Disorder Association (NEDA) announced that it was temporarily shutting down a hotline chatbot named Tessa – intended to replace human staff – because it was giving weight loss advice to people with eating disorders. A August report published by the Center for Countering Hate (CCDH) found that AI chatbots generated harmful eating disorder content 23% of the time. In 94% of this harmful content, the bot warned that the advice was dangerous, even though all the information was still readily available.
When Reverse asked ChatGPT to create a low-calorie meal plan for weight loss, the chatbot dutifully stuck to three meals and two snacks. The meals, although consisting of standard healthy fare like oatmeal in the morning and chicken fillet with broccoli and quinoa for dinner, were quite bland and limited. ChatGPT makes no mention of an individual’s dietary preferences and does not provide instructions on alternatives, such as plant-based protein options for vegetarians. The snacks were also pretty pitiful: carrot sticks with hummus and exactly 10 almonds.
The author asked about calorie restrictive meal plans for weight loss. ChatGPT automatically defaulted to a diet of 1,200 calories per day for women and 1,500 calories per day for men.
Screenshot by Miriam Fauzia
But what was most concerning was that ChatGPT defaulted to 1,200 calories per day for an adult woman and 1,500 calories for men. Debbie Fetterassistant professor with a doctorate in nutritional biology at the University of California, Davis, says Reverse this type of calorie restriction is unhealthy: it can trick the body into thinking it is going through a famine and, therefore, “giving up its metabolism to conserve energy.” If you’re hoping to lose weight on this diet, you might experience the exact opposite.
“There are a lot of disordered eating habits prevalent on social media where people can fall into something and not realize that in reality the (eating) habits are very limiting and restrictive in nature,” says Fetter . “It’s scary not knowing what the source of this information is.”
Is there a way to use ChatGPT responsibly?
No one should blindly listen to ChatGPT’s “health” recommendations, Fetter and Malone say. But they also say it’s still possible to integrate chatbot into your nutrition planning as long as a professional is involved, such as a dietitian.
Professional help is essential to create a nutrition plan that takes into account more individual variables such as an individual’s physical lifestyle, personal goals, and food preferences based on what they actually like eating or its cultural values.
“It’s scary not knowing what the source of this information is.”
“(A registered dietitian) would provide some general information: find out your goals, if there are any health issues or food allergies, or if a person is looking to change their health to a certain extent, such as their blood cholesterol levels or their body composition,” Fetter says.
According to Malone, considering a person’s nutritional needs holistically rather than through the limited and biased lens of a chatbot is the true value of a human connection.
“In any health care profession, we have a role of clinical judgment, which is not necessarily an answer,” she says. “You have to take into account all these factors about the situation, the person, their history, and make a decision based on the medical evidence available and what makes sense. I think it would be almost impossible for a computer.
Although there may be a day when computers can do everything – help us manage our lives in a way that preserves our well-being with a simple string of code – we clearly haven’t reached that point yet.
